michaelni changed the topic of #ffmpeg-devel to: Welcome to the FFmpeg development channel | Questions about using FFmpeg or developing with libav* libs should be asked in #ffmpeg | This channel is publicly logged | FFmpeg 7.0.1 has been released! | Please read ffmpeg.org/developer.html#Code-of-conduct
rellla has quit [Remote host closed the connection]
System_Error has quit [Remote host closed the connection]
System_Error has joined #ffmpeg-devel
<BBB>
Sean_McG: I think any is fine?
<Sean_McG>
yeah it's fine.. I just don't know why my simple flacdsp_wasted32 does not work
iive has quit [Quit: They came for me...]
<BBB>
feel free to upload code somewhere, show checkasm output and ask for comments
<cone-708>
ffmpeg David Rosca master:d2d911eb9a2f: lavc/vaapi_av1: Avoid sending the same slice buffer multiple times
<cone-708>
ffmpeg David Rosca master:fe9d889dcd79: lavc/vaapi_decode: Make it possible to send multiple slice params buffers
<cone-708>
ffmpeg David Rosca master:510494760cde: lavc/vaapi_h264: Fix merging fields in DPB with missing references
AbleBacon has quit [Read error: Connection reset by peer]
ssfl has quit [Remote host closed the connection]
kurosu has joined #ffmpeg-devel
HarshK23 has joined #ffmpeg-devel
DeHackEd has quit [Remote host closed the connection]
DeHackEd has joined #ffmpeg-devel
<frankplow>
I also get a 500 error trying to access the Git repo
<JEEB>
git.videolan.org seems to work OK
<JEEB>
at least the webui, that is
<frankplow>
Yeah git.videolan.org seems fine. Webui for git.ffmpeg.org works, but actual repo doesn't.
Livio has joined #ffmpeg-devel
cone-708 has quit [Quit: transmission timeout]
Livio has quit [Ping timeout: 256 seconds]
Livio has joined #ffmpeg-devel
Krowl has joined #ffmpeg-devel
_whitelogger has joined #ffmpeg-devel
jarthur has quit [Quit: jarthur]
Livio has quit [Ping timeout: 256 seconds]
Krowl has quit [Read error: Connection reset by peer]
rvalue has quit [Read error: Connection reset by peer]
rvalue has joined #ffmpeg-devel
Krowl has joined #ffmpeg-devel
<kurosu>
Random question a bit out of place, but what does the browser landscape looks like in terms of full range? I know the GFX driver will do whatever it wants, but still, are they still struggling ?
<kurosu>
(I see issues that are 5-6yo, but not much after that)
System_Error has joined #ffmpeg-devel
System_Error has quit [Remote host closed the connection]
IndecisiveTurtle has quit [Remote host closed the connection]
IndecisiveTurtle has joined #ffmpeg-devel
System_Error has joined #ffmpeg-devel
MetaNova has quit [Remote host closed the connection]
MetaNova has joined #ffmpeg-devel
mkver has joined #ffmpeg-devel
IndecisiveTurtle has quit [Ping timeout: 255 seconds]
jamrial has joined #ffmpeg-devel
Teukka has quit [Read error: Connection reset by peer]
novaphoenix has quit [Quit: i quit]
novaphoenix has joined #ffmpeg-devel
Teukka has joined #ffmpeg-devel
Teukka has quit [Changing host]
Teukka has joined #ffmpeg-devel
MetaNova has quit [Remote host closed the connection]
MetaNova has joined #ffmpeg-devel
lexano has joined #ffmpeg-devel
Krowl has quit [Read error: Connection reset by peer]
Krowl has joined #ffmpeg-devel
cone-593 has joined #ffmpeg-devel
<cone-593>
ffmpeg Stone Chen master:55e9c758f07d: libavcode/x86/vvc: change label to vvc_sad_16 to reflect block sizes
<cone-593>
ffmpeg Stone Chen master:d82c50355587: tests/checkasm/vvc_mc: for SAD, only test valid subblock sizes
System_Error has quit [Remote host closed the connection]
<cone-593>
ffmpeg Rémi Denis-Courmont master:c53d42380d3f: lavc/lpc: optimise RVV vector type for compute_autocorr
<cone-593>
ffmpeg Rémi Denis-Courmont master:91b5ea7bb930: lavc/vp8dsp: R-V V vp8_luma_dc_wht
<cone-593>
ffmpeg Rémi Denis-Courmont master:fd39997f7278: lavc/vp7dsp: add R-V V vp7_luma_dc_wht
<cone-593>
ffmpeg Rémi Denis-Courmont master:fa3b153cb1eb: lavc/vp7dsp: R-V V vp7_idct_add
blb has quit [Ping timeout: 268 seconds]
blb has joined #ffmpeg-devel
System_Error has joined #ffmpeg-devel
rvalue has quit [Read error: Connection reset by peer]
rvalue has joined #ffmpeg-devel
ccawley2011 has joined #ffmpeg-devel
tufei_ has joined #ffmpeg-devel
tufei has quit [Ping timeout: 260 seconds]
tufei_ has quit [Ping timeout: 260 seconds]
AbleBacon has joined #ffmpeg-devel
IndecisiveTurtle has joined #ffmpeg-devel
Krowl has quit [Read error: Connection reset by peer]
tr4nq_ has joined #ffmpeg-devel
<tr4nq_>
ohi
<kepstin>
kurosu: last i checked, at least with h264, you could expect browsers to ignore full range indications and always display as limited range :/
<kurosu>
-_-
<tr4nq_>
does chrome/chromium use MPEG-5 Essential Video Coding at all?
<Lynne>
firefox specifically does it correctly, as the file is from an issue they had open about them not dealing with it
<kurosu>
Lynne: how would you check a browser does it correctly from this? I imagine a screen capture would have gone through potentially several layers of postprocessing? Also depending on whether HW decode happens (and so what postprocessing was involved)
<kurosu>
Would you remember keywords of said issue so that I can search it myself (unless you remember easily its URL)
<JEEB>
ah, so they took trees and larger block sizes frmo newer formats
<JEEB>
and then binary arithmetic coding from JPEG
<JEEB>
yea, so it might be better than AVC, but it conveniently forgets that there are royalty free formats around that have been developed already
<tr4nq_>
i'm quite new to the world of codecs so i'll have a steep learning curve but lucky for me i like coding and learning new things
<JEEB>
during HEVC times there was VP9, and then now there's AV1 since 2018
<tr4nq_>
it's a codec jungle out there
<JEEB>
in general there's two trees that don't really cross-pollinate. there's the ISO/IEC (MPEG) tree of formats, and then the "geez look at the licensing mess" Google and later AOM format tree
<tr4nq_>
i see
<JEEB>
MPEG seems to be really hardcore on "don't bring up licensing during format development"
<JEEB>
so their formats generally end up with some vague FRAND notices and that's it
<JEEB>
as we could see with HEVC, its usage outside of piracy ended up being: broadcast, UHD blu-ray, apple
<JEEB>
VVC seems to follow the same trend
<tr4nq_>
cool
<JEEB>
meanwhile streaming related companies are moving more and more towards the modern royalty-free formats
<tr4nq_>
yeah why wouldn't they
<JEEB>
development wise the difference is fun between MPEG and AOM
<tr4nq_>
how so
<JEEB>
even I could register to various workgroups' mailing lists
<JEEB>
at MPEG
<JEEB>
I was on the HEVC development ML for years
<JEEB>
and the documents are public!
<tr4nq_>
\o/
<JEEB>
meanwhile AOM is a cabal that keeps its documents at lock until the format is ready
<tr4nq_>
gotcha
<JEEB>
which is really funny since AOM formats are generally called "free" or "open"
<JEEB>
their development method most definitely is not :D
<kurosu>
EVC to me is Samsung deciding neither HEVC nor AV1 suited them. Nonetheless, Qualcomm and Tencent contributed to it, and there are I think 2 profiles, one wanting to be RF (not sure how much it resembles AVC), and the other RAND. Not sure anything is achieved by that. If anything, there are more companies agreeing to the AoM license
<tr4nq_>
i had no idea the world of codecs involves that level of politics
<kurosu>
I still don't understand why they did it, why they even pushed it in ffmpeg. Sounds like Google and VP8/VCB, and see their fate
<kurosu>
Codec licensing is a multi-billion business
<tr4nq_>
TIL
<JEEB>
also checking JVET-experts actually made me notice some interesting'ish stuff
<JEEB>
> Video CICP problem reports and proposed actions
<JEEB>
> CICP enum for generic subsampling and sample location
<kurosu>
> MPEG convenor moving from Microsoft to Dolby
<JEEB>
yea
<JEEB>
was fun seeing that name with a different ()
<JEEB>
although his email in the doc is @outlook.com
<JEEB>
lul, this now 100% explains why H.273 2023-09 is still not out
<JEEB>
> H.273 V3 was Consented in July 2023 and has been pre-published. However, it remains pending final publication due to its informative reference to SMPTE ST 2128, since that standard remains unpublished.
<JEEB>
> The reference to SMPTE ST 2128 is found in clause 8.3, Table 4 – Interpretation of matrix coefficients (MatrixCoefficients) value, in the row for MatrixCoefficients equal to 15 (IPT-C2). Since this reference is merely informative, the reference could simply be deleted, although this would result in an absence of an explanation of why this colour representation scheme is specified in H.273.
<JEEB>
ebin
<JEEB>
indeed it would result in an absence of an explanation of why that value exists
<JEEB>
for those who don't recall, value 15 is what A**le forced D to specify as they apparently disliked that DoVi profile 5 content was marked "unspecified" for everything
<JEEB>
and then > [AHG9/CICP] On constraints and reference of MatrixCoefficients
<JEEB>
bytedance seems to have noticed that some constraints had slipped when moving stuff from H.264/5 to H.273
cone-593 has quit [Quit: transmission timeout]
Livio has joined #ffmpeg-devel
Livio has quit [Ping timeout: 252 seconds]
Livio has joined #ffmpeg-devel
Sean_McG has joined #ffmpeg-devel
BtbN has quit [Remote host closed the connection]
BtbN has joined #ffmpeg-devel
Krowl has quit [Read error: Connection reset by peer]
<BBB>
that vvc simd is nerd-sniping me
<BBB>
it's so hard to resist
<BBB>
if you unpcklqdq and hqdq the same register, you're just splitting it in "low half" and "high half". if you then unpcklwd it, you could just as well have unpckl/hwd'ed it to begin with, making the unpacklqdq unnecessary
<BBB>
and unpack-to-word (assuming these are unsigned integers and m14 is zero), addd, packssdw is just a saturated add. use paddsw
<Lynne>
JEEB: could you tell me what the consensus was about the channel layout bikeshedding?
<Lynne>
I can't tell, and marton hasn't responded
cworley has quit [Quit: bye]
bilboed has quit [Quit: Ping timeout (120 seconds)]
bilboed has joined #ffmpeg-devel
cworley has joined #ffmpeg-devel
<BBB>
kurosu: I always read evc as mpeg members embrace-and-extending royalty-free codecs (vp9/av1) after fearing it might become succesful
<BBB>
but I'm a bit biased probably
IndecisiveTurtle has quit [Ping timeout: 268 seconds]
<Lynne>
yeah, mpeg is an organization that cannot survive without patent licenses
<kurosu>
Seems the EVC goes beyond NIH into general distrust attitude
<kurosu>
EVC case
<Son_Goku>
I'm pretty sure EVC exists solely because of broadcast standards orgs *requiring* an MPEG standard codec
<kurosu>
Well, that would need to be incorporated into an application standard such as ETSI or ATSC, so... Hasn't happened, will not happen?
System_Error has quit [Remote host closed the connection]
<BBB>
that's a lot of letters for a standard nobody needed or asked for
wyatt8740 has quit [Remote host closed the connection]
System_Error has joined #ffmpeg-devel
wyatt8740 has joined #ffmpeg-devel
Livio has quit [Read error: Connection reset by peer]
Livio has joined #ffmpeg-devel
<JEEB>
Lynne: basically I would prefer if at least new things would follow at least one of the specs they base on, while Marton seems to prefer some sort of continuation of other existing values. I am not gonna fight on my side, but then it becomes even more important to have those matching standardized names in the post-name `///< ` comment or so :P I think that's where it ended up.
System_Error has quit [Remote host closed the connection]
<JEEB>
would be nice if someone had a copy of IEC 62574
<cosminaught>
it looks like one could purchase a copy of it. How helpful would it be?
System_Error has joined #ffmpeg-devel
<JEEB>
so for the +/-90 degree side surrounds it would be `///< Lss - Left side surround, SiL - Side left` (and same with R/Right)
<JEEB>
for the top apparently +/-110 degree ones... `///< Lvs - Left vertical surround, TpLS - Top left surround`, although the latter is not verified since it doesn't appear in the ITU BS.2051 doc :P
<JEEB>
cosminaught: it would be useful, but in general we don't support organizations putting such stuff behind paywalls
<JEEB>
BS.2051 has U+110 as Ltr in some layout, but then D and ISO specs on hand refer to TpLs, so it seems to exist :P
<cosminaught>
JEEB: I agree that standards shouldn't be behind paywalls, but at the same time if you need to answer some important questions and this standard would help I'm happy to buy a copy and answer the question
<Sean_McG>
I've started getting 500 Error from git.ffmpeg.org when trying to update, anyone else seeing this?
IndecisiveTurtle has joined #ffmpeg-devel
<JEEB>
Sean_McG: yea apparently that host isn't too healthy, git.videolan.org is OK tho
<courmisch>
I need to report an impediment to scrum master
<JEEB>
:D
<courmisch>
I'm going to merge that sbrdsp hack why elenrils are not looking
System_Error has quit [Remote host closed the connection]
System_Error has joined #ffmpeg-devel
<thardin>
git.ffmpeg.org broken?
<JEEB>
yea, git.videolan.org seems fine though
<thardin>
ssh works
<thardin>
good enough for rebasing
IndecisiveTurtle has quit [Remote host closed the connection]
Livio has quit [Ping timeout: 256 seconds]
ccawley2011 has quit [Read error: Connection reset by peer]
<ePirat>
Is the ffmpeg trac down`
<ePirat>
*`
<ePirat>
oh no its just incredibly slow
dionisis has joined #ffmpeg-devel
<michaelni>
thardin, it seems the latest security update to git broke it. I dont fully understand it, i hope one of the other root admins will take a look and have some more insight
<michaelni>
it should only affect public git, developer ssh access to all servers should be fine
<michaelni>
that is git over ssh
kurosu has quit [Quit: Connection closed for inactivity]
<michaelni>
gitweb also seems fine
<michaelni>
btw i do understand why it produces a 500 error but it doesnt work even when i fix that (i did that for https://git.ffmpeg.org/michael.git and that doesnt work but i cannot find any errors on the server)
<michaelni>
if someone has some ideas or wants to help ...
<michaelni>
that is once the 500 is fixed (which is fixed by adding the repository to safe.directory) it still doesnt work but no more errors show up
<michaelni>
debuging this is not my area, i think its more efficient if i spend my time working on ffmpeg that produces more return per hour ...
<michaelni>
i could downgrade git but thats kind of not great or i could try to upgrade to latest git and forget about the official ubuntu package but i really dont want to experiment around on the server. So all feedback is welcome here !
<michaelni>
of course i have 0 clue if upgrading git would fix this
tufei has joined #ffmpeg-devel
<thardin>
yeah ssh is fine
lexano has quit [Ping timeout: 268 seconds]
<ePirat>
michaelni, for michael.git I just get no data from the server it seems…
<BtbN>
michaelni: from my understanding, you need to run the add safe-directory command _as apache user_, and not touch the ffmpeg git config at all